大数据环境下多媒体网络舆情信息的语义识别与危机响应研究

?In the big data environment of network, with the popularity of various new network businesses and the strengthening of citizens' awareness of participation, the complexity of diffusion and diversification of network public opinion has reached an unprecedented level.The semantic recognition of multimedia online public opinion is not only the real demand for monitoring and analyzing the public opinion crisis，but also the key issues which should be studied and solved for promoting the assessment and response ability of public opinion crisis. In recent years the relevant researches mainly surrounded the public opinion dissemination and coping mechanisms, these achievements greatly promoted the development process of public opinion crisis management in our country, but still failed to achieve the ideal solution for the mass and isomerism public opinion mining and standardized mechanism for assessment and response.so facing a dilemma of lacking the method of administration public opinion in big data environment. In view of this, the project will start from the perspective of multimedia online public opinion in big data environment,do a deep research on the solution of multimedia online public opinion identification and assessment on semantic.Then the project analyze the new connotation of multimedia online public opinion in big data environment,constructing the system architecture of public opinion knowledge base,completing the function of concurrent access to heterogeneous public opinion;construing the semantic identification and mining model of multimedia online public opinion based on semantic grid and concept lattice,exploration multidimensional fusion method from semantic vector to assessment index,and establishing the mechanism of public opinion crisis response.

在网络资源大数据化的环境下，随着各种网络新业务的普及和公民参与意识的加强，网络舆情产生扩散的复杂性和多元化已达到前所未有的程度。针对多媒体网络舆情信息的语义识别既是分析监测舆情危机的现实需求，也是提升舆情危机评估与响应能力所必须解决的关键问题。近年来相关研究主要围绕着舆情传播与应对机制，其成果虽然促进了我国网络舆情危机管理的现实推进，然而基于语义的海量异构舆情信息整合挖掘与规范化评估响应机制建设方面始终未能求得理想的解决方案，因而面临着大数据化舆情信息管理手段匮乏的困境。 本项目从多媒体网络舆情的大数据化出发，研究多媒体网络舆情语义识别与评估的解决方案。解析"大数据化多媒体网络舆情"新内涵，构建舆情知识库的体系架构，实现异构舆情信息的并发获取功能；基于语义网格、概念格技术构建多媒体网络舆情语义识别与分析模型，建立语义特征向量到评估指标体系的多维融合，并建立相应的舆情危机响应机制。

大数据环境下多媒体网络舆情的语义识别与危机响应是清朗网络空间、规范网络管理的重要内容。近年来我国网络舆情分析与治理的研究成果虽然极大的促进了网络舆情监管工作的推进，但面对多种媒体形式的、大数据化的网络舆情信息分析、舆情危机应对等工作中，仍存在着响应体系不完善，监管手段匮乏等困境。有鉴于此，本项目基于目前国内外的语义处理技术，研究大数据化多媒体网络舆情的传播特征、演变规律、以及舆情危机的形成机理与应急响应范式。主要研究内容包括：①重构了大数据化多媒体网络舆情信息的演进与传播模式；②构建出以管理和复用大数据化多媒体网络舆情案例为目标的知识库群；③提出大数据化多媒体网络舆情信息的并发获取模型；④提出异构的多媒体网络舆情信息语义识别的方案；⑤形成一套评估大数据化多媒体网络舆情信息危机等级的指标体系；⑥提出了以大数据化多媒体网络舆情危机分型为基础的响应机制。